Re: A proposal to remove xsd:ENTITY, xsd:ID, and xsd:IDREF datatypes from OWL 2

Thanks Boris.

Just for the records, as I will not be at the telco next week (19th): I
agree with your proposal of dumping them.

Ivan

Boris Motik wrote:
> Hello,
> 
> Recently, Jos de Bruijn has made a comment that the xsd:ENTITY, xsd:ID, and xsd:IDREF datatypes are not recommended for use in RDF
> and OWL 1. Therefore, he asked us why we included these datatypes into OWL 2. His original e-mail is here:
> 
> http://lists.w3.org/Archives/Public/public-owl-comments/2008Oct/0000.html
> 
> I looked into this and here are my findings. The xsd:ENTITY datatype has been derived from xsd:NCName by placing a rather strange
> restriction on its value space:
> 
>  The value space of ENTITY is scoped to a specific instance document.
> 
> This probably means you cannot define the value space globally, but need to somehow associate it with the document where xsd:ENTITY
> is being used. This is clearly quite complex and has not been intended in my original definition of the OWL 2 datatype map.
> 
> The xsd:ID and xsd:IDREF datatypes exhibit no such problems: their value space is exactly the same as for xsd:NCName. Therefore, I
> am not really sure why RDF and OWL 1 warned against their usage.
> 
> 
> Either, I believe that the simplest way to address Jos's comment is to delete all three datatypes from the rdf:text and OWL 2
> specifications. Since neither of these datatypes provides us with something more than xsd:NCName, we are not really losing any
> expressivity here.
> 
> Regards,
> 
>  Boris
> 
> 

-- 

Ivan Herman, W3C Semantic Web Activity Lead
Home: http://www.w3.org/People/Ivan/
PGP Key: http://www.ivan-herman.net/pgpkey.html
FOAF: http://www.ivan-herman.net/foaf.rdf